There is only a single data stream for all the compressed tables, but the code was originally written for separately compressed tables then retrofitted, so this is an attempt to streamline it a bit.